VABS Manual
VABS Manual
VABS Manual
Wenbin Yu
February 13, 2013
Introduction
VABS (Variational Asymptotic Beam Sectional Analysis) is a code implementing the various beam
theories1, 2, 3, 4, 5, 6, 7, 8 based on the concept of simplifying the original nonlinear three-dimensional
(3D) analysis of slender structures into a two-dimensional (2D) cross-sectional analysis and a onedimensional (1D) nonlinear beam analysis using a powerful mathematical method, the variational
asymptotic method.9 VABS takes a finite element mesh of the cross section including all the details of geometry and material as inputs to calculate the sectional properties including structural
properties and inertial properties. These properties are needed for the 1D beam analysis to predict
the global behavior of the slender structure. The 3D pointwise displacement/strain/stress distribution within the structure can also be recovered based on the global behavior of the 1D beam analysis.
Since most of the theoretical details are presented in pertinent papers and collected in the book
by Prof. Hodges,10 this manual will only serve to help readers get started using VABS to solve
their own composite beam problems. This manual addresses the history of the code, its features,
functionalities, conventions, inputs, outputs, maintenance, and tech support.
VABS History
The research project that gave birth to VABS was initiated by Prof. Dewey Hodges when he was
first introduced to the variational asymptotic method by Prof. Berdickevsky at Georgia Tech in 1989
and has been ongoing ever since till the time of writing. The program name VABS first appeared
in [11]. The original version of VABS was a research code written in Fortran 77 by Prof. Cesnik.
Although the current version of VABS is a completely rewritten code, it is important to note that
the original version laid an able foundation for the present code and facilitated its development.
The fall semester of 1998, when Prof. Yu began his graduate study at Georgia Tech, marked the beginning of the transition of VABS from a research code to a production design and analysis tool for
VABS is copyrighted and commercialized by Utah State University Technology Commercialization Office and
Georgia Institute of Technology Research Cooperation. All rights reserved.
Associate Professor, Department of Mechanical and Aerospace, Engineering, Utah State University, Logan, Utah
84322-4130.
practicing engineers. The code was rewritten from scratch in Fortran 90/95, with all unnecessary
restrictions eliminated, and the computing and memory efficiency greatly enhanced. At the same
time, Prof. Censik was continuing his work on VABS for piezoelectric materials at MIT and later
at University of Michigan. And Prof. Yu continues his work on VABS for multiphysics modeling
at Utah State University. For this reason, there are two variants of VABS: the Georiga Tech/Utah
State VABS, released and maintained by Profs. Yu and Hodges, and UM/VABS, released and
maintained by Prof. Cesnik. From henceforth in this manual the term VABS will refer only to
the Georgia Tech/Utah State VABS, and what follows is only applicable to this code. Many researchers and engineers all over the world are actively using VABS which is becoming the standard
tool for design and analysis of composite slender structures such as helicopter rotor blades, wind
turbine blades, high aspect ratio wings, composite bridges, and other slender structural components.
VABS II was released in June 2004, with the major enhancement to remove the need of asking
the user to choose arbitrary point constraints and let the code determine the singularity and apply
the corresponding constraints. Other improvements of VABS II include calculation of principal
inertial axes, the generalized mass matrix, and neutral axes, and significant reduction of the computing time for large size problems.
The main feature of VABS 3.7 is to carry out the recovery up to the second order which provides
a better prediction for the 3D recovered fields in comparison to known exact solutions.
The main feature of VABS 3.6 is to use an improved method for optimizing the numbering of finite
element mesh. For large problems, it is much faster that the method used in previous versions.
And also most of the time, VABS 3.6 can handle larger problems. And also the most recent version
of gfortran compiler is used for compiling the code resulting in much faster recovery.
The main feature of VABs 3.5 is for oblique cross-sectional analysis, the inputs are given in the
oblique cross-sectional system while in previous versions, the inputs are given in the normal crosssectional coordinates. Although starting VABS 3.5, users can use long names of input file, including
spaces in the path and file names.
The main new feature is for expanding g in the modeling for initially curved/twisted beams. For
some cases such a change made significant differences for obtaining first and second correction of the
stiffness matrix due to initial curvature and twist. Such a change is verified using an initially curved
strip for which an elasticity solution is obtainable. The input file for this case is isorectTrif2.dat.
A 64-bit version of VABS is also available since VABS 3.4.
The main new feature is introducing a new input format, and keeping the previous input format
as optional. In the new format, the user only needs to provide one real number for 1 as few users
take advantage of the nine real numbers for 1 , which is useful for elements with highly curved
edges. In the new format, we introduce layer definition so that a layer type instead of material
type is provided for each element. Each layer is defined through a unique combination of material
type and layerup angle 3 . It is more economical than assigning 3 for each element, as what we
have done in previous format, because the number of layers usually is much less than the number of
elements. These two changes reduce approximately 3/4 of real numbers needed for VABS inputs,
saving space and time. These changes will also simplify the development of VABS preprocessors as
it is easier to calculate just one number for 1 for each element and programs flows better with the
newly introduced layer definitions.
VABS was originally designed to run as a stand alone code and its error handling, memory
allocation/de-allocation, and I/O were handled with this use in mind. However, in recent years,
more and more users began to explore the possibility of using VABS in a design environment. This
motivates the major upgrade of VABS to VABS III through restructuring the code.
Since the first release of VABS III, a few users have asked the difference between VABS III and
previous versions, in particular VABS 2.1.1 which was the last release and the code accompanying
Prof. Hodges book.10 Overall, VABS III is a much improved code in both accuracy and efficiency.
The main difference can be described according to the following two aspects.
As far as functionalities concerned, VABS III
1. Uses the correct constraints so that it can reproduce the 3D elasticity theory for isotropic
prismatic beams. This change affects the warping functions, and affects all stiffness
models except the classical one. Such a correction enables VABS to reproduce the 3D
elasticity theory for isotropic prismatic beams and thus enables VABS to provide a better
modeling for prismatic or initially curved/twisted composite beams (VABS 3.0).
2. Recovers 3D stress/strain fields at each node in addition to the Gauss points. The
recovered 3D stress/strain fields are expressed in both the beam coordinate system and
the material coordinate system. VABS 2.1.1 only recovers 3D stress/strain fields at the
Gauss points expressed in the beam coordinate system. For visualization, nodal values
are convenient. To apply failure criterion, values expressed in the material coordinate
system are necessary (VABS 3.0).
3. Handles isotropic, orthotropic, and anisotropic material differently. Previous versions
treat all materials as orthotropic and only and must take a total of 9 elastic constants.
4.
5.
6.
7.
VABS III allows general anisotropic material with as many as 21 elastic constants and
isotropic materials with as few as 2 elastic constants (VABS 3.0).
Can model hygrothermal effects of composite beams due to temperature and moisture
changes. As a companion capability, VABS Conduction is developed to carry out a dimensional reduction for the 3D conduction problem. VABS Conduction can be requested
separately (VABS 3.1).
Updates the transformation procedure into the Timoshenko model from the asymptotic
energy. A new perturbation method is developed to capture the effects due to initial
curvatures/twist during the transformation. The prediction for Timoshenko stiffness is
generally improved, even for some prismatic beams (VABS 3.2).
Outputs the average of 3D stresses/strains within each element for convenience of postprocessing (VABS 3.2.2).
Provides an option for recovering the 3D displacement/strain/stress fields based on linear
beam theory (VABS 3.2.4).
Since January 2006, VABS has been commercialized by Utah State University along with Georgia
Institute of Technology. This creates difficulty for some VABS users, such as students using VABS
for their studies or engineers using VABS for not-work-related problems. To meet this need, we are
committed to develop VABS Lite for each new release as freeware available to all interested users.
VABS Lite is a limited version of VABS which is capable of classical and Timoshenko modeling of
prismatic beams. VABS Lite can model cross section with up to 500 nodes and whatever number
of elements formed by the limited number of nodes.
10
VABS Features
Hence along with the features of previous versions, the most recent version of VABS has the
following features:
1. It is a highly modularized code written in Fortran 90/95. All the problem-dependent arrays
are allocated dynamically during run time, and the user can use all the memory up to the
limit of the machine. All the outstanding abilities of array handling in Fortran 90/95 have
been exploited.
2. It adopts highly efficient techniques to reduce the requirement of RAM and increase the
computational efficiency. Now cross sections as complex as real composite rotor blades with
hundreds of layers can be easily handled by a laptop computer.
3. It has a general element library that includes all the typical 2D elements such as 3, 4, 5,
6-noded triangular elements and 4, 5, 6, 7, 8, 9-noded quadrilateral elements. Users are free
to choose the type of elements, and different types of elements can be mixed within one mesh,
if necessary.
4. It can deal with arbitrary layups. Users can provide one parameter for the layup orientation
and one parameter for the ply orientation to uniquely specify the material system in the
global coordinate system. Nine parameters can be used for the ply orientation if a ply is
highly curved and the ply angle is not uniform within an element.
5. It detects singularities and properly removes them to render the solution as a true representation of the theory. Older versions before VABS II dealt with them approximately by asking
the users to input four constraints on three distinct, user-specified nodes. The arbitrariness of
the older approach can affect the refined models, and sometimes may even render the linear
system unsolvable.
6. It applies the four constraints on the warping functions in such a way that the 3D elasticity solution can be reproduced for isotropic beams, correcting a mistake related with these
constraints in previous versions.
7. It does not require the beam reference line to be the locus of cross-sectional area centroids.
VABS can calculate the centroid for any arbitrary cross section, and users can choose their
own reference line for the convenience of the 1D global beam analysis.
8. It can deal with isotropic materials, orthotropic materials, and general anisotropic materials,
while all the old versions treat all materials as orthotropic.
9. It can be quickly and conveniently integrated with other environments such as computeraided design environments, multidisciplinary optimization environments, or commercial finite
element packages.
11
VABS Functionalities
VABS is a versatile cross-sectional analysis tool and can obtain all the useful information related
with beam sections. Specifically, VABS has the following functionalities:
1. Calculate the 66 cross-sectional mass matrix, written in terms of the mass per unit length,
and the first and second distributed mass moments of inertia about the three axes. Based on
the information provided by the mass matrix, VABS calculates the cross-sectional center of
mass, the inertial principal axes, the principal mass moments of inertia, and the mass-weighted
radius of gyration.
2. Construct the classical model, represented by a 44 stiffness matrix for prismatic or initially
curved/twisted composite beams with normal or oblique cross sections. Based on the classical
stiffness model, VABS can calculate the location of neutral axes, or so-called tension center.
3. Construct the generalized Timoshenko model, represented by a 66 stiffness matrix for prismatic or initially curved/twisted composite beams with normal cross sections. From the
generalized Timoshenko model, VABS can calculate the shear center (or elastic axis) location
for arbitrary sections.
4. Construct the generalized Vlasov model, represented by a 55 stiffness matrix for prismatic
or initially curved/twisted composite beams with normal cross sections, which is important
for thin-walled beams with open sections.
5. Incorporate the trapeze effect, a nonlinear effect in the 1D constitutive law that stems from
3D geometrical nonlinearity, a very important effect for beams under large centrifugal forces.
The composite beam could be either prismatic or initially twisted and curved.
6. Recover 3D pointwise displacement, strain, and stress fields for the global behavior of a 1D
global beam analysis using the classical model, the generalized Timoshenko model, or the
generalized Vlasov model. Multiple recovery runs can be performed for different inputs of
stress resultants without repeating the cross-sectional analysis. The recovered stress/strain
fields are evaluated both at the nodal positions and Gaussian points. They are expressed in
both the material coordinate system and the beam coordinate system.
7. Can model hygrothermal effects of composite beams due to temperature and moisture changes.
As a companion capability, VABS Conduction is developed to carry out a dimensional reduction for the 3D conduction problem, which can be requested separately.
x3
x3
x1
x2
12
VABS Conventions
To understand the inputs and interpret outputs of the program correctly, we need to explain some
conventions used by VABS.
First, VABS uses a right hand system, the beam coordinate system, denoted as x1 , x2 and x3 ,
where x1 is along the beam axis and x2 and x3 are the local Cartesian coordinates of the cross
section, see Figure 1 for a beam with an arbitrary cross section. Usually, for rotor blades, x1 is
along the direction of the span and points to the tip, x2 is along the direction of the leading-edge
to the trailing edge and points to the direction of the leading edge, and x3 can be determined
by the right hand rule. Usually the origin of x1 is located at the root of the blade, yet the user
is free to choose the origin of x2 and x3 at an arbitrary point of the cross section, such as the
mass center, centroid, or shear center. Detailed information is needed to define the cross-sectional
geometric domain spanned by x2 and x3 and the materials that occupy that domain. Also, certain
characteristics along the span direction, such as initial curvature/twist or taper, are needed for
cross-sectional analyses when they are not equal to zero. The obliqueness should be specified when
the angle between x1 and the x2 -x3 plane is not equal to 90 , that is, when reference cross section
is not normal to the reference line, such as the case of a swept wing. It is noted that the beam
coordinate system is the same as the undeformed beam coordinate system b defined in Ref. [10].
Second, VABS numbers the nodes of each element in the counterclockwise direction, as shown
in Figure 2 for triangular elements and Figure 3 for quadrilateral elements. Nodes 1, 2, and 3 of the
triangular elements and nodes 1, 2, 3, and 4 of the quadrilateral elements are at the corners. Nodes
5, 6, 7 of the triangular elements and nodes 5, 6, 7, 8, 9 for quadrilateral elements are optional nodes.
The recovered 3D displacements are values at each node expressed in the VABS beam coordinate system (Figure 1). However, stresses and strains are most accurately evaluated at Gaussian
3
3
6
1
1
1
4
5
2
2
Figure 2: VABS triangular element node numbering and corresponding integration schemes
4
3
7
8
3
4
59
1
1
1
6
2
Figure 3: VABS quadrilateral element node numbering and corresponding integration schemes
integration points. Gaussian integration schemes for different orders of the two types of elements
are also shown in Figures 2 and 3. The red interior points correspond to the integration scheme
for linear elements and the green interior points correspond to the integration scheme for quadratic
elements. VABS can also recover 3D stresses and strains at each node as suggested by our industry
users. The recovered stresses, and strains are expressed in both the beam coordinate system and
the material coordinate system, which is essential for applying failure criteria.
VABS allows users to use various kinds of units. However, it is necessary to be absolutely
consistent in the choice of units to avoid errors. Particularly, users must never use the pound as a
unit of mass to avoid confusion. When pounds are used for force and feet for length, the unit of
mass must be the slug = lb-sec2 /ft; if inches are used for length along with pounds for force, then
the unit of mass must be lb-sec2 /in.
Finally, to understand the VABS input convention for composite layups, we need to find relationships among three coordinate systems: the beam coordinate system (x1 , x2 , x3 ) used by the
user to define the geometry, the material system (e1 , e2 , e3 ) used by the user to define the material
properties, and an intermediate one to define the ply plane (y1 , y2 , y3 ). As shown in Figure 4, the
13
VABS Inputs
Although a few preprocessors, such as PreVABS, have been developed to create VABS input files,
it is still beneficial for advanced users, particularly those who want to embedded VABS in their
familiar environment, to understand the meaning of the input data.
The first line lists two newly introduced integer flags arranged as: format flag nlayer. If the
first flag is 1, the input is prepared in the new format, otherwise, it is prepared in the old format.
The second integer provides the number of layers in the section. Note, here layer is defined as a
unique combination of material type and layup orientation, it does not necessarily corresponds to
the definition used in the manufacturing sense. For example, even if a section is made of a single
isotropic material, we consider it has one layer. Hence, nlayer should be always given a value
greater than one if format flag=1. Note nlayer is not used in the old format.
The second line has four flags arranged as: Timoshenko flag recover flag thermal flag. The
first flag can be only 1 or 0 and the second flag can be 0, 1, or 2. Their uses are explained in the
following:
If Timoshenko flag is 1, VABS will construct both the classical model and the generalized
Timoshenko model. If it is 0, it will only construct the classical model.
y3 , e3
y2
y2
y3 , e3
y3 , e3
y2
y2
y3 , e3
10
x3 , y3
y2
x2
y1
x1
normal cross section
11
number of nodes, nelem the total number of elements, and nmate the total number of material types.
The next nnode lines are the coordinates for each node arranged as: node no x2 x3 , where
node no is an integer representing the unique number assigned to each node and x2 , x3 are two real
numbers describing the location (x2 , x3 ) of the node. Although the arrangement of node no is
not necessary to be consecutive, every node starting from 1 to nnode should be present.
The next nelem lines list 10 integers for the nodes for each element (also known as the connectivity relations). They are arranged as: elem no node 1 node 2 node 3 node 4 node 5 node 6
node 7 node 8 node 9, where elem no is the number of element and node i (i = 1, 2, . . . , 9) are
nodes of this element. If a node is not present in the element, the value is 0. If node 4 is 0, the element is a triangular element; see Figures 2 and 3 for the VABS numbering convention. Although
the arrangement of elem no is not necessary to be consecutive, every element starting
from 1 to nelem should be present.
If f ormat I = 1, that is, if the new format is used, the next nelem lines list the layer type
and the layer plane angle (1 ) for each element as: elem no layer type 1 , where layer type is an
integer representing which layer the element elem no belongs to, and 1 is a real number describing
the layer plane angle for the element. 1 can be calculated at any material point belonging to the
element, such as the centroid. Although the arrangement of elem no is not necessary to
be consecutive, every element starting from 1 to nelem should be present. For isotropic
materials, 1 will not enter the calculations.
If f ormat I is not equal to one, that is, if the old format is used, the next nelem lines list
the material type and layup parameters for each element as: elem no material type 3 1 (9), where
material type is an integer representing the type of the material for the element elem no, 3 is a
real number representing the layup angle in degrees for this element, and 1 (9) is an array storing
nine real numbers for the layer plane angles at the nodes of this element. For simplification, if the
ply orientation can be considered as uniform for this element, 1 (1) stores the layer plane angle
and 1 (2) = 540 , and all the rest can be zeros or other real numbers because they do not enter the
calculation. If the element has fewer than nine nodes, zeros are to be input for the corresponding
missing nodes, as in the case for connectivity. Although the arrangement of elem no is not
necessary to be consecutive, every element starting from 1 to nelem should be present.
For isotropic materials, neither 3 nor 1 (9) will enter the calculations.
If f ormat I = 1, that is, if the new format is used, the next nlayer lines define the layers used
in the section. They are arranged as:
layer id mate type 3
where layer id is an integer denoting the identification number for each layer, mate type is an integer denoting the material type used in the layer, and 3 is a real number denoting the layup
orientation. For example, if layer 1 is made of material 1 and having 15 layup, we will provide
the information as 1 1 15.0.
The next nmate blocks defines the material properties. They are arranged as:
12
mat id orth
const1 const2 ....
where mat id is the number of material type, orth is the flag to indicate whether the material is
isotropic (0), orthotropic (1) or general anisotropic (2). The rest are material constants.
For isotropic materials, orth is 0, if thermal flag is 0, there are 3 constants arranged as:
E
where E is the Youngs modulus, is the Poissons ratio, and is the density of the material.
Poissons ratio must be greater than -1.0 and less than 0.5 for linearly elastic isotropic materials,
although VABS allows users to input values that are very close to those limits.
If thermal flag is 3 and orth is 0, and there are 4 constants arranged as:
E
including the Youngs moduli (E1 , E2 , and E3 ), the shear moduli (G12 , G13 , and G23 ), the Poissons
ratios (12 , 13 , and 23 ), and the mass density (). The convention of values is such that these
values will be used to form the following generalized Hookes law for composite materials:
11
1/E1
0
0
21 /E2
0
31 /E3
11
2
0
1/G
0
0
0
0
12
12
12
213
0
0
1/G
0
0
0
13
13
=
12 /E1
22
0
0
1/E2
0
32 /E3
22
2
0
0
0
0
1/G
0
23
23
23
33
13 /E1
0
0
23 /E2
0
1/E3
33
If thermal flag is 3 and orth is 1, and there are 13 constants arranged as:
E1 E2 E3
G12 G13 G23
12 13 23
11 22 33
where 11 , 22 , 33 are the coefficients of thermal expansion along three directions.
The material constants are expressed in the material coordinate system; see Figure 4. It is also
emphasized that if the users are provided material properties in a different coordinate system, or
the arrangement of stresses and strains are different from what VABS uses, proper transformation
of the material properties is needed.
13
For general anisotropic materials, orth is 2, if thermal flag is 0, there are 22 constants arranged
as:
c11 c12 c13 c14
c22 c23 c24
c33 c34
c44
c15
c25
c35
c45
c55
c16
c26
c36
c46
c56
c66
These values are defined using the following generalized Hookes law:
11
c11 c12 c13 c14 c15 c16
11
2
12
12
13
c
c
c
c
c
c
2
13
23
33
34
35
36
13
=
c14 c24 c34 c44 c45 c46 22
22
23
223
33
c16 c26 c36 c46 c56 c66
33
If thermal flag is
c11 c12 c13 c14
c22 c23 c24
c33 c34
c44
14
If recover flag is 1, the user needs to provide additional information obtained from the 1D global
beam analysis in the same input file. To carry out recovery based on the classical model, VABS
requires the following data:
u1 u2 u3
C11 C12 C13
C21 C22 C23
C31 C32 C33
F1 M1 M2 M3
where u1 , u2 , and u3 are the 1D beam displacements along x1 , x2 , x3 , respectively. The matrix Cij ,
with i = 1, 2, 3 and j = 1, 2, 3, is the direction cosine matrix defined as
Bi = Ci1 b1 + Ci2 b2 + Ci3 b3 with i = 1, 2, 3
where B1 , B2 , and B3 are the base vectors of the deformed triad and b1 , b2 , and b3 are the base
vectors of the undeformed triad. Details of this definition can be found in Ref. [10]. ui and Cij
are needed only for recovering 3D displacements. If the user is not interested in 3D displacements,
these values can be arbitrary real numbers. F1 is the sectional axial force, M1 is the sectional
torque, M2 is the sectional bending moment around x2 , and M3 is the sectional bending moment
around x3 . The sectional stress and moment resultants are needed for recovering 3D stress and
strain distribution within the cross section. For example, if the user only wants to see the pointwise
stresses or strains under 1 unit compressive axial force along with 1 unit bending moment around
x2 , the inputs can be arranged as:
0 0 0
1 0 0
0 1 0
0 0 1
1 0 1 0
To carry out recovery based on the generalized Timoshenko model, VABS requires the following
data:
u1 u2 u3
C11 C12 C13
C21 C22 C23
C31 C32 C33
F1 M1 M2 M3
F2 F3
f1
f2
f3 m1 m2 m3
f10
f20
f30 m01 m02 m03
f100 f200 f300 m001 m002 m003
f1000 f2000 f3000 m000
m000
m000
1
2
3
where the additional data F2 and F3 are sectional transverse shear forces along x2 and x3 , respectively. f1 , f2 , f3 are distributed forces (including both applied forces and inertial forces) per unit
span along x1 , x2 , x3 respectively. m1 , m2 , m3 are distributed moment (including both applied and
inertial moments) per unit span along x1 , x2 , x3 respectively. The prime denotes derivative with
15
x1 ,
()00 =
2
,
x21
and ()000 =
3
.
x31
To carry out recovery based on the generalized Vlasov model, VABS requires the following data:
u1 u2 u3
C11 C12 C13
C21 C22 C23
C31 C32 C33
11
1
2
3
01
001
000
1
where 11 is the beam axial stretching strain measure,
1 is the twist measure,
2 and
3 are the
curvature measures around x2 and x3 respectively. It is noted that the global behavior needed for
recovery should not violate the small strain assumption. Otherwise, you might get some unexpected
results. For example, if your transverse shear stiffness is around 2.5 N., then input a shear force
resultant of 1 N might be too large as the shear strain is 0.4, which cannot be considered as small.
The input file should be ended with a blank line to avoid any possible incompatibility of different
computer systems. The input file can be given any name as long as the total number of the
characters of the name including extension is not more than 256. You are suggested to use a
unique extension say vab for you to identify such files with VABS. For the convenience of the user
to identify mistakes in the input file, all the inputs are echoed in a file named input file name.ech.
Error messages are also written at the end of input file name.ech.
14
VABS Outputs
Sectional properties obtained by VABS are stored in input file name.K. Some results are listed
as scalars, and some are listed as matrices. Typical scalar results include the mass center, the
principal axes, the geometric center, the neutral axes (or tension center), and the elastic axis (or
shear center). Typical matrix results include the cross-sectional mass and stiffness matrices of
the classical model, the cross-sectional stiffness matrix of the generalized Timoshenko model, the
cross-sectional stiffness matrix of the generalized Vlasov model, and the cross-sectional stiffness
matrix considering the trapeze effect. The corresponding flexibility matrices are also provided in
the output file. These values are all expressed with respect to the beam coordinate system chosen
by the user except when the Vlasov flag is equal to 1, the sectional values are expressed with respect
to the shear center, although the shear center is still expressed with respect to the user coordinate
system. The definitions of these properties are briefly summarized here for the convenience of end
users. For more details, please refer to VABS related publications.
The elements of the mass matrix are arranged as
0
0
0
xm3 xm2
0
0
xm3
0
0
0
0
xm2
0
0
0
xm3 xm2 i22 + i33
0
0
xm3
0
0
0
i22
i23
xm2
0
0
0
i23
i33
16
where is mass per unit length, (xm2 , xm3 ) is the location of mass center, i22 is the mass moment
of inertia about x2 axis, i33 is the mass moment of inertia about x3 axis, i23 is the product of
inertia. VABS also calculates the mass matrix at the mass center, and the principal mass moments
of inertia and principal inertial axes following standard formulas in textbooks on dynamics. With
the knowledge of mass center and principal inertial axes, VABS outputs the mass per unit span,
mass moments of inertia about x1 , x2 , x3 , and mass-weighted radius of gyration, the square of which
is the sum of the two in-plane cross-sectional mass moments of inertia (which just happens to be
the mass moment of inertia around x1 ) divided by the mass per unit span.
The geometric center of the cross section is defined as:
Xg2 =
where < >=
integration.
RR
< x2 >
<1>
Xg3 =
< x3 >
<1>
dx2 dx3 . Please noted that only the geometry occupied by a material enters the
The 44 stiffness matrix is used to expressed the following constitutive relation for the classical
beam model:
1
11
22 S23 S24
M1
S
S
12
1
1
S
M
S13 S23 S33 S34
M3
S14 S24 S34 S44
3
The neutral axes or the tension center (Xt2 , Xt3 ) is defined as the point of the cross section through
which no bending deformation will occur when an axial force applied at this point. The stiffness
constants Sij depend on the choice of the beam coordinate system, initial curvature/twist, as well
as the geometry and material of the cross section. For prismatic beams made of isotropic material
with the beam reference line located at the tension center and x2 and x3 aligned with the principal
bending axes, the stiffness matrix is a diagonal matrix as follows:
EA 0
0
0
0 GJ
0
0
0
0 EI2
0
0
0
0 EI3
the diagonal terms of which are the well known engineering beam constants. For prismatic beams
made of isotropic material with the beam reference line not at the tension center and x2 and x3
not aligned with the principal bending axes, the stiffness matrix will look like:
EA
0
S3
S2
0
GJ
0
0
S3
0
EI2
EI23
S2 0 EI23
EI3
with S =< Ex > and EI23 =< Ex2 x3 >.
17
T , a 4 1
If thermal flag is equal to 3, VABS also output 1D nonmechanical stress resultants (N
matrix) and thermal strains (
T , a 4 1 matrix) due to temperature for the classical beam model
as
F1
11
M1
1
T
= S
N
M2
M3
3
T .
and T = S1 N
The 6 6 stiffness matrix is
alized Timoshenko model:
F1
S11
F
S12
F3
S13
=
S14
M2
S15
M3
S16
S13
S23
S33
S34
S35
S36
S14
S24
S34
S44
S45
S46
S15
S25
S35
S45
S55
S56
S16
S26
S36
S46
S56
S66
11
212
213
1
2
3
11
12
213
=S
1
2
3
where 212 and 213 are the engineering transverse shear strains. The generalized shear center
or elastic axis (Xs2 , Xs3 ) defined as the point of the cross section through which no torsion will
occur directly due to the sectional transverse shear forces. It is noted that the barred quantities
= ()|2 =2 =0 .
11 ,
1,
2,
3 relate to their unbarred counterparts as ()
12
13
For initially curved and twisted beams, there are two 4 4 stiffness matrix for the classical
model. The first classical stiffness matrix corresponds to the zeroth-order strain energy. The second classical stiffness matrix (correct up to the second order) corresponds to the second-order strain
energy with respect to the small parameter related with initial twist/curvatures.
The 5 5 stiffness matrix is used to expressed the
alized Vlasov model:
S11 S12 S13
F
M1
S12 S22 S23
M2
= S13 S23 S33
M3
M
S15 S25 S35
S15
S25
S35
S45
S55
11
01
where M is so-called bi-moment corresponding to the twist rate 01 . S55 is the commonly called
restrained warping constant, although the complete 5 5 matrix is needed for composite beams
with significant restrained warping effects. For thin-walled beams with open sections, it is meaningful to have a generalized Vlasov model based on choosing shear center as its reference. Hence to
obtain a generalized Vlasov model, VABS finds the shear center first and then shifts the origin of
the coordinate system to the shear center and calculate the 55 stiffness matrix. It is noted that
bars or hats are put over the stiffness constants for different models to indicate the fact that they
18
11
12
1 213
U=
1
2
2
3
11
1
+
S11
S12
S13
S14
S15
S16
S12
S22
S23
S24
S25
S26
S13
S23
S33
S34
S35
S36
S14
S24
S34
S44
S45
S46
(11 A + 1 B1 + 2 C2
S16
11
S26
2
12
S36
2
13
S46
S56
2
S66
3
11
1
+ 3 D3 )
S15
S25
S35
S45
S55
S56
19
15
VABS Installation
16
Prof. Yu is committed to maintaining and providing tech support for VABS. A Google group is
specifically set up for information exchange related with VABS. Users are highly encouraged to
sign up through http://groups.google.com/group/hifi-comp to receive most recent news of VABS,
ask questions, and share with others. A technical question should be posted in the Google
20
group before it will be answered. A page of VABS FAQ will be constantly updated in the
group. Before you ask questions, please do the following:
1. Read the VABS manual carefully, if you have not done so;
2. Check the error message at the end of input file name.ech;
3. Make sure that you have provided the right input data through input file name.ech, which is
VABSs understanding of your input file;
4. Check the VABS FAQ page on the Google group;
5. Post your question in the discussion section of the group.
A web site (hifi-comp.com) has also been created to provide more exposure of the codes Prof.
Yu has developed including VABS, VAPAS, and VAMUCH through Internet. .
17
Epilogue
After a period of continuous development spanning more than 15 years, VABS has reached a
level of maturity, and its accuracy has been extensively verified by its developers and users. The
performance and robustness of code have been continuously improved based on feedback from its
users throughout the world. Although VABS has been designed in such a way that end users do not
have to fully understand its theoretical foundation (the details of which are spelled out in VABS
related publications), further questions are inevitable because VABS represents a new paradigm to
analyze composite beams which is drastically different from other methods. Nevertheless, it should
be clear that VABS is the best available code for engineers to design and analyze composite beam
cross sections.
References
[1] W. Yu, D. H. Hodges, V. V. Volovoi, and C. E. S. Cesnik. On Timoshenko-like modeling of
initially curved and twisted composite beams. International Journal of Solids and Structures,
39(19):5101 5121, 2002.
[2] W. Yu, V. V. Volovoi, D. H. Hodges, and X. Hong. Validation of the variational asymptotic
beam sectional analysis. AIAA Journal, 40(10):2105 2113, Oct. 2002.
[3] W. Yu and D. H. Hodges. Elasticity solutions versus asymptotic sectional analysis of homogeneous, isotropic, prismatic beams. Journal of Applied Mechanics, 71(1):15 23, 2004.
[4] W. Yu and D. H. Hodges. Generalized Timoshenko theory of the variational asymptotic beam
sectional analysis. Journal of the American Helicopter Society, 50(1):46 55, 2005.
[5] W. Yu, D. H. Hodges, V. V. Volovoi, and D. F. Eduardo. A generalized Vlasov theory of
composite beams. Thin-Walled Structures, 43(9):1493 1511, 2005.
21
[6] D. H. Hodges and W. Yu. A rigorous, engineering-friendly approach for modeling realistic,
composite rotor blades. Wind Energy, 10(2):179193, 2007.
[7] S. Roy, W. Yu, and D. Han. An asymptotically correct classical model for smart beams.
International Journal of Solids and Structures, 44(25-26):84248439, 2007.
[8] D. Han, W. Yu, and S. Roy. A geometrically exact active beam theory for multibody dynamics
simulation. Smart Materials and Structures, 17(4):11361147, 2007.
[9] V. L. Berdichevsky. Variational-asymptotic method of constructing a theory of shells. PMM,
43(4):664 687, 1979.
[10] Dewey H. Hodges. Nonlinear Composite Beam Theory for Engineers. AIAA, Washington DC,
2006.
[11] D. H. Hodges, A. R. Atlgan, C. E. S. Cesnik, and M. V. Fulton. On a simplified strain energy
function for geometrically nonlinear behaviour of anisotropic beams. Composites Engineering,
2(5 7):513 526, 1992.
22